EVENT SUMMARY ¦ Credit and Collections Technology Think Tank: Spring 2025

Remember to take part:
The 2025 Spring Benchmarking Study

Another great day yesterday at the Credit Connect Think Tank.
Many thanks to the Credit Connect team for organising such a fabulous event during the day and evening. Loads of ideas and new ways of thinking… a couple of highlights
➡️ How do we use AI… not to find a needle in the haystack, but to remove some of the hay
➡️ The importance of focusing on good customer outcomes and being able to evidence this… rather than just sticking to a process
➡️ FOS volume, the thinking around fee economics and what this means for the future
➡️ Looking at changes… and the very act of applying for credit is an indicator that something is potentially going on in a customer’s life
➡️ There is only one customer… with lots of relationships… inc public and private.. we need to coordinate the approach
➡️ AI continues to move at pace… lots ahead live in processes already, more on the agent-customer interaction to come soon
Also don’t forget to take part in the spring-summer Collections Benchmarking study. If would like to take part click on the QR code/link here.

Key takeaways from each session

Session 1: Credit and Collections Risk Challenges

Separation of Customer Base: The industry is seeing a split between financially secure and financially stretched customers, requiring more nuanced strategies.
Changing Credit Behaviour: Customers are increasingly using credit for essentials rather than discretionary spending, altering repayment patterns.
Regulatory Focus on Vulnerability: The FCA remains heavily focused on consumer vulnerability, pushing firms to tailor solutions.
Affordability Under Scrutiny: Affordability assessments are mandated, but panellists question their effectiveness as a sole metric for creditworthiness.
Personalisation is Key: Customised communication and solutions are essential for effectively managing vulnerable and mainstream customer segments.
Technology Enables Flexibility: Firms need advanced tooling and data analytics to respond to the complexity of modern customer financial behaviour.
Data Needs Are Growing: Traditional credit scoring is no longer sufficient—open banking and behavioural data are becoming more relevant.
Consumer Expectations and Education: Misunderstandings around credit impacts require better customer education to manage false perceptions and complaints.
Weaponisation of Complaints: Regulatory frameworks and compensation schemes are at risk of being exploited, complicating risk management.
Regulatory Complexity is Increasing: Upcoming obligations such as product sales data and board reports require granular and transparent reporting.
Sectoral Learnings: Experiences from motor finance and high-cost credit offer cautionary tales for wider financial services regarding redress and oversight.
Balancing Commercial Viability and Compliance: Firms must walk a fine line between regulatory expectations and maintaining commercial agility.

New Ideas

Tailored Engagement Models: Emerging approaches focus on real-time, reactive personalisation of services based on granular behavioural data.
Dynamic Affordability Assessments: Innovations in affordability monitoring using live financial data, rather than static snapshots.
Risk Modelling with Behavioural Data: Beyond credit scores, firms are exploring social and transactional data to build more accurate risk profiles.
Predictive Regulatory Compliance: Enhanced reporting tools allow for pre-emptive identification of poor outcomes, improving compliance.

Key Statistics

20 million: Estimated number of potentially vulnerable customers in the UK.
6 million: Number of people reportedly avoiding credit applications due to fear of rejection.
70%: Referenced improvement in affordability-related governance through revised credit reporting.
50%: Proportion of the population potentially classified as vulnerable.

Session 2: Assessing Affordability and Customer Vulnerability

50% Vulnerability Claim Discussion: Claims that half of all customers are vulnerable requires nuanced understanding—vulnerability is multifaceted, not binary.
Dynamic Journeys Needed: Customer journeys should be redesigned to accommodate fluid, context-driven vulnerability and affordability situations.
Support Needs > Labels: Shifting focus from vulnerability labels to specific support needs is seen as a more mature and effective approach.
Behaviour-Driven Engagement: Real behavioural insights are key to triggering appropriate interventions at the right moments in the customer lifecycle.
Do No Harm Principle: A recurring theme is to avoid worsening a customer’s situation through poor engagement or rigid procedures.
Early Intervention Essential: Effective use of data and signals can enable earlier and more appropriate support offers.
Data Sharing Challenges: Persistent barriers exist around cross-sectoral and intra-sectoral data sharing, inhibiting holistic support.
Empowering Frontline Staff: High-performing agents given autonomy and better training deliver improved outcomes without additional cost.
AI as a Tool, Not a Threat: AI and chatbots show promise for empathetic, anonymised engagement, particularly with younger consumers.
Shift to Long-Term Volatility: Rising costs, tax burdens, and macroeconomic shifts suggest volatility is the “new normal.”
Sector Collaboration is Critical: Central and local government, utilities, and financial services must align on vulnerability protocols and definitions.
Balance Between Automation and Human Touch: Automation is valuable for scale, but must be balanced with skilled human engagement for complex, sensitive situations.

New Ideas

Anonymised AI Counselling: Large language models are increasingly being used as private, judgment-free sources of financial and emotional support.
Dynamic Support Triggers: Behavioural and payment pattern analytics are being explored to proactively flag support needs before crises escalate.
Flexible Journey Design: Movement away from fixed flags towards dynamic, need-based journey structures for customer service and collections.
High-Skill Collector Models: Pilots with well-compensated, empowered agents show better customer outcomes and comparable financial performance.

Key Statistics

50%: Estimated proportion of vulnerable customers, frequently referenced but viewed with caution due to definitional issues.
£60,000: Annual salary of a high-performing collections agent in an experimental model delivering strong outcomes.
10.4 million: People reported to face mental health challenges affecting financial vulnerability.
6 months: Suggested timeline for more flexible, practical data sharing frameworks.

Session 3: Maximising Customer Engagement

Customer Engagement Has Evolved: There’s a shift from traditional, channel-led strategies to psychologically-informed, behaviour-led models.
Digital-First Design is Now Core: Many lenders are prioritising digital-first journeys with self-service capabilities and dynamic communication flows.
Psychological Trust Building is Vital: Emotional triggers and trust signals drive engagement more than transactional efficiency alone.
Human + Digital Blending is Essential: Optimal engagement relies on the right balance between digital tools and human intervention at key points.
Social Media as a Discovery and Engagement Tool: Platforms like TikTok and Reddit are being used to monitor sentiment, drive traffic, and shape customer journeys.
Behavioural Science Shapes Contact Strategy: Nudges, content framing, and multi-touchpoint strategies are being adopted to influence and support customer behaviour.
Digital Accessibility is Non-Negotiable: Compliance with accessibility standards (e.g. WCAG 2.1) is becoming a core requirement across all digital services.
Personalisation Beyond Demographics: Engagement strategies increasingly adapt to customer personas, behaviour, and neurodiversity.
Self-Serve Must Support the Right Outcomes: Measuring and ensuring that self-service platforms deliver effective and fair results is critical.
AI and Automation for Scale and Insight: Firms are investing in AI to personalise messaging, assess engagement patterns, and improve operational efficiency.
New Metrics for Engagement Required: Traditional metrics like average handle time are less relevant in a digital-first, hybrid environment.
Cybersecurity and Customer Awareness Growing: Rising concerns about phishing and data security require firms to be vigilant and proactive.

New Ideas

Integrated Social Listening: Use of platforms like TikTok and Reddit to gather real-time sentiment and engagement signals.
Behavioural Nudges in Digital Journeys: Applying principles from behavioural science to content design, digital journeys, and user prompts.
WCAG-Compliant Customer Journeys: End-to-end commitment to accessible design, backed by external audits.
Persona-Led Engagement Models: Real-time customer data and AI-informed personas guiding communication tone, timing, and medium.
AI-Driven Comms Optimisation: Use of models to tailor and refine message content at scale based on response analysis.

Key Statistics

25%: Proportion of customers using credit score tracking tools.
3 days: Average advance notice given for credit score changes.
99%: Digital accessibility compliance target.

Session 4: The impact of AI in Credit and Collections

AI’s Role is Expanding Across Functions: From conversational interfaces to process automation, AI is increasingly embedded across collections and customer service.
AI Must Be Purpose-Led: The panel emphasised that AI should be tied directly to clear business goals, not adopted simply because it is trending.
Conversational AI Outpacing Basic Chatbots: Transitioning from static bots to dynamic, NLP-driven virtual assistants enables better engagement and self-service.
Hyper-Personalisation at Scale is Now Feasible: AI enables real-time adjustments to tone, content, and timing of communications at an individual level.
Maintaining Human Oversight Remains Crucial: AI should not operate in isolation—human-in-the-loop models are needed for complex or sensitive decisions.
Agent Roles Are Shifting: Entry-level roles are vanishing, replaced by positions that require high emotional intelligence and specialist training.
False Positives Are a Key Risk: AI must be carefully monitored to avoid incorrect vulnerability or risk classifications that could lead to inappropriate responses.
Legacy Systems Create Friction: Integration with legacy technology remains a barrier to realising AI’s full potential in collections operations.
FCA Sandbox Seen as Key Enabler: The FCA’s sandbox provides a safe space for testing AI innovation within regulatory boundaries.
Model Transparency is Essential: Financial institutions need to understand and explain AI model outputs, especially when customer decisions are impacted.
Data Use Must be Ethical and Proportionate: Sentiment analysis, usage patterns, and voice data must be handled with transparency and restraint.
Skills Development is Urgent: The industry must upskill and retrain staff to navigate the AI-enabled environment, balancing automation with empathy.

New Ideas

Intelligent Virtual Assistants: Evolution from rule-based bots to AI-driven assistants capable of interpreting complex language and context.
AI-Assisted Quality Monitoring: Automated review of customer calls to flag vulnerabilities, dissatisfaction, and potential harm.
Micro-Template Messaging Systems: AI generates personalised communications dynamically based on behaviour, time, and customer context.
AI-Aided Sentiment and Intent Detection: Speech and behaviour analytics identify distress, urgency, or miscommunication in real-time.
FCA Sandbox Collaboration: Enables AI solution testing under live conditions with regulatory support and guidance.

Key Statistics

38%: Predicted proportion of organisations expected to use AI in daily operations within ten years.
78%: AI response accuracy rate in one implementation discussed, covering customer interaction handling.

RO-AR insider newsletter

Receive notifications of new RO-AR content notifications: Also subscribe here - unsubscribe anytime